Long Tailed Duck Release Candidate

43 views
Skip to first unread message

Johann Koenig

unread,
Dec 29, 2016, 7:38:55 PM12/29/16
to codec...@webmproject.org
Long Tailed Duck has branched from master and is now available on the
'longtailedduck' branch in the libvpx repository.

Please test with your systems and let us know via the bug tracker[0]
or mailing lists[1] if you encounter any issues. We hope to finalize
this release in January.

What's new?
- General speed improvements to VP9.
- In the highbitdepth (HBD) configuration, encoding and decoding of 8
bit content is no longer appreciably slower than the non-HBD
configuration. This allows deploying a HBD capable library without
impacting existing 8 bit users.
- Fuzzing fixes.

You can get the code with:
$ git clone https://chromium.googlesource.com/webm/libvpx
$ cd libvpx
$ git checkout longtailedduck

[0] https://bugs.chromium.org/p/webm/issues/list
[1] https://groups.google.com/a/webmproject.org/forum/#!forum/codec-devel

Johann Koenig

unread,
Jan 9, 2017, 6:24:34 PM1/9/17
to codec...@webmproject.org
We will be rolling up to at least
cabc29ba24b0948439b48ce49ef8693f8d37b3cd to pick up some key changes:
mips/dspr test coverage and idct fixes:
Add mips dspr2 partial idct tests
Fix mips dspr2 idct32x32 functions for large coefficient input
Fix mips dspr2 idct16x16 functions for large coefficient input
Fix mips dspr2 idct8x8 functions for large coefficient input
Fix mips dspr2 idct4x4 functions for large coefficient input
Add mips dspr2 vp9 intrapred tests

Various vp9 realtime improvements:
vp9: 1 pass cbr mode: increase threshold for gf_cbr_boost_pct usage.
vp9: Enable more aggresive short circuit for speed 8.
vp9: Compute source sad for every superblock when partition copy is on.
vp9: Set short circuit to level 3 for VGA for speed 8.
vp9: 1 pass cbr: allow noise estimation down to 360p.
vp9: 1 pass vbr: Skip find_predictors in pickmode when source is altref.

neon cleanups:
Clean DC only idct NEON intrinsics
Clean idct 8x8 neon functions
Add high bitdepth 8x8 idct NEON intrinsics

Other fixes:
Fix for out of range motion vector bug in joint motion search
Make sub-pixel mv search's return value consistent with the return type
Bug fix to avoid random crashes during ARNR filtering
Reply all
Reply to author
Forward
0 new messages